Automated Progressive Delivery Using GitOps and Service Mesh
summary
Intuit shows an automated progressive delivery pipeline combining Argo Rollouts with an Envoy/istio service mesh, deployed across ~230 clusters and >7 million pods (peak >1M TPS). Claims: service-to-service latency drops ~30%, certificate rotation shrinks from 90 days manual to hours automated, and a canary that returns HTTP 500 triggers automatic rollback in demo. Engineering details: multi-cluster discovery required admiral endpoint sync, Argo CD manages configs, ~2,000 legacy services need migration, and metric aggregation and ML-based anomaly detection are next.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:59Why service mesh at intuit?
- 8:07Cross-cluster service discovery
- 10:19Progressive Delivery at Intuit
- 14:15Issues uncovered
- 15:41Onboarding 2000 services x2
- 17:22Configuring analysis templates
- 18:52Feature investment and research
related talks
35mProgressive Delivery with Argo Rollouts
46mProgressive Delivery in the Kubernetes era
28mProgressive Delivery With Argo Rollouts and Kubernetes Gateway API for Multi-Directional...
43mProgressive Application Delivery: Istio Ambient Mesh and Argo Rollout in Action- 43mProgressive Application Delivery: Istio Ambient Mesh and Argo Rollout in Action
30mProgressive Delivery Made Easy With Argo Rollouts
40mExpanding Horizons: Leveraging Advanced Deployment...
35mLeveling Up Your CD: Unlocking Progressive Delivery on Kubernetes